• aditya mehrotra.

more design practice and learning: designing a spaceship [updates]

Updated: Apr 25, 2020

So I found this: https://ntrs.nasa.gov/search.jsp?N=4294964234

And this: https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20110023499.pdf#page28


This is the server for NASA's technical reports and it also contains the safety factors NASA uses in their designs. So there's some really good points in here such as (in paraphrase) "just because you're designing for SPACE doesn't mean you should only design for space, structures should not yield on earth during transportation."


OTHER FACTORS TO CONSIDER:

Fatigue life: should be designed for 4.0x the service life

Buckling: any structures subject to compressive/in-plane shear stresses should undergo buckling review. ESPECIALLY thin-wall devices.

High-cycle fatigue life: anything subject to high-cycle fatigue should have a minimum high-cycle fatigue life of 10x the service life.

Creep Avoidance/Creep Life: we need to avoid any form of creep as it could lead to rupturing or collapse, minimum creep life should be 4.0 times the service life.


FACTORS OF SAFETY:

NASA requires that all designs be tested physically and compared to simulation results.

So here's the new challenge. We need to learn a little about fatigue, high-cycle fatigue, buckling, and brush up on creep. And then we can get to things like designing the craft. Our thinking was a little too limited last time and that was not a proper structural analysis of what is happening in the mars lander.


Let's look at some online resources here:

"Verification of SKYCRANE using..."

https://ntrs.nasa.gov/search.jsp?R=20150008792&hterms=skycrane+mars&qs=Ntx%3Dmode%2520matchall%7Cmode%2520matchall%26Ntk%3DAll%7CAll%26N%3D4294964234%26No%3D10%26Ntt%3Dmars%7Cskycrane


So this is going to give us a much better idea of the requirements that go into making something like this and how to analyze something like it. Then we'll take a look at these things:

BUCKLING:

- https://www.continuummechanics.org/columnbuckling.html

FAITUGE:

- http://web.mit.edu/course/3/3.11/www/modules/fatigue.pdf

HIGH CYCLE FATIGUE:

- https://www.sciencedirect.com/topics/engineering/high-cycle-fatigue

CREEP

- https://www.sciencedirect.com/topics/engineering/high-cycle-fatigue


^We're going to go through and take notes on all of the above resources. I'm pretty sure that the creep avoidance just means literally we need to design so the material never enters the creep region (that's the viscoelastic region) and the plastic deformation region. But I want to make sure I know what I'm talking about. I don't have much to do this weekend, so over the next few days I'm going to spend some time each day going through things like the topics above to gain more knowledge on what these topics are - and through the process we will design this landing frame. So I feel like a logical order to look at this in is:


(1) look at the design requirements of the MSL sky crane in the PDF to see what kinds of loads and forces the frame has to take and then TRANSLATE them to the scale of chip.

(2) do a standard load FEA on that structure/hand-calcs on that structure until, for example, we have the safety factors we want (listed above). So get a general frame design you know.

(3) then go to buckling analysis of the frame to see if anything will buckle with the thin walls.

(4) then see about fatigue

(5) then see about creep


That's the plan...

So first, the free body diagram of the system and starting the design all over again. Honestly, the diagram is not too different from this. We have an accurate FBD, but I want to consult with a MechE friend on how to analyze it.

In the meantime. I'm going to start the CAD of an initial design with constrains and with parameters so we can actually draw a better diagram. My friend who's a MechE suggested a dynamic simulation in fusion not the static one which will allow us to generate the forces more accurately.


The first thing we're going to change from the previous design is we're going to have the rover sit under a BOX frame. Not just a + sign on top in the box. The box will give the system more support. I'm trying to fix the constraints right now. It's taking a while to cad because of the constraints. I'm trying to do CAD properly.

Here is the new, fully-defined 3D sketch of the lander. Note the design differences between this and the previous. There's an upper "box" instead of just a frame (should have known). We're also going to trace this thing with the same ID/OD tube structure we did last time but in a smarter way. Let's get started on that since that might take some time.

So the "smarter" way is we're sweeping each "bar" individually after the 6 MAIN SQUARES that make up the frame for two reasons. (1) That's how it would likely be built, and (2) it provides for better JOINTS between all the bars (more accurate joints that resemble how it would be built).

Here's after most of the key sweeps have been handled. We see the five major frame components (the central box, and the four outlying boxes).


Also... after sweeping half the frame I did learn fusion has a pipe tool that I should probably use next time. We will use that next time for now the frame is fine!

Yay! we have a frame. Now let's change the material to titanium and chill a little for now because that took about two hours.

The ID = 0.75 here and the OD = 1.00 for now. We're going to fix all this later in FEA. We should have limited meshing issues because of how this thing was created (much better than the last frame). If the circle tube doesn't work we can even try things like square tube/etc.


How would we make this? We'd water jet the flat stock and we'd cut and cope the tubes and weld the whole thing together. It really wouldn't be that hard to do. If we're looking for easier assembly, we'd need to re-design the frame a lot.

With some finishing touches. We're now ready to move onto things like FEA and start refining the design. We're mainly concerned with changing the ID and OD of the titanium pipes as well as the structure of the frame itself so the frame is as light as possible but it can bear the structural loads required of it. We're going to use the same technique of crating a structural member to apply the forces to.

So fusion 360 has things like buckling study and etc... We're going to try a buckling study here to see what actually happens. Then we can try things like event study or static study. We're going to apply gravity as it would be to the frame, and we will apply a downwards force of 500N on the frame in the circular location to symbolize it holding the rover's weight. Then 800N of upwards thrust on all four of the rockets will be generated.

What my MechE friend suggested was to create a PLANE above the structure that limits the Y-movement so that we don't need to define any part of the structure as completely fixed. We place a plane above it and just say "no penetration" or something, or we take the top of the frame and fix it in the Y direction only. This allows the study to run without adding too many extra loads to the structure. There will be some added load from the fixing, but hopefully because of the spread of the load it won't affect the study too much. Same material as yesterday ID-0.75, OD-1.00! We're going to use the adaptive meshing this time (the percentage) not the absolute 10mm. We think the errors have been fixed. Let's let it mesh for a bit.


Again the studies we will be doing on this frame on the computer will be things like bucking and static stress simulation studies to find the values of forces and etc. Then we'll be doing hand calc to verify/etc, and then look at things like fatigue and stress and stuff through cyclic loads.


Question: Does fatigue play a role here? YES ABSOLUTELY:

Think of it like this. We're currently simulating a static upwards thrust of 800N on each of the motors. But really what's going to happen isn't that all 4 of the engines will fire at a constant force for the entirety of the descent. They'll cyclically fire.

FULL

OFF

FULL

OFF

FULL

OFF

To slow the craft down first. And then the will probably cyclically fire for longer to slow further. Until we hit the powered descent stage where things will be like "300N per thruster constant." Or something.


So here are the studies we're going to do:

(1) The buckling study with the parameters above. Let's see if the structure buckles due to its thin walls --> we'll verify this with a hand-calc.

(2) The static stress study --> with the SAME parameters as above, and we'll verify this as well with a hand-calc.

(3) The fatigue study with cyclic loads as described. We'll design that better later. BUT IT'S IMPORTANT WE DO IT IN THIS CASE THE LOADS ARE CYCLIC.


How long is this thing supposed to last? Exactly one trip that's it that's all its needed for. So when doing cyclic, let's not overdo it. We're not trying to re-design a NASA craft here, I'm trying to learn. We're going to start with static stress, then go to buckling, then go to cyclic.


This video is also probably worth looking into. As I said, the FBD of this structure and that of a quadcopter is not that different. The only difference is the size and what is propelling it.


Anyways - back to our simulation we're going to generate a mesh of 5mm! Ooh! 5mm mesh worked! Let's try the static sim.


Okay that filed because the system was not constrained enough. So we're going to make that plane fixed in both directions and re-run it. So yea, there goes our brilliant planar-constraint idea.

Okay so here are the results form the simulation. So the minimum and maximum safety factor in this case is like 15 which is way too high. We want to keep the frame light while also making the frame strong right. So this frame would survive but it's heavier than it needs to be so let's try again! We're going to just run structural simulations first to bring down the safety factor before we do any other analysis.


For the next run we're going with a 0.75" outer tube diameter and a 0.075" wall thickness. Again we're going to use a 5mm mesh size. We're using the same exact loads and constraints.

We're getting this thin wall error again which is good which means we can and should do a buckling analysis on the structure. We'll do that after we bring our safety factor down past like 15 oops. Maybe we don't need titanium. Or maybe we just have too many bars. For certain, our titanium does NOT have to be this thick. I really hope this meshing works. We might need to run these sims in solidworks to like get accurate results. We're going to make the ID 0.5 to see if we can avoid a meshing error again.


Also another thing I just thought of, is that we know, from the simulation, the maximum stress, we can try to find out what direction that's in, and we can use that to start doing some hand calc on what we think the diameter of the tube should be. We're going to use a simplified model of this ridiculous structure. We did a 2-D simplification and here are the results.

So what this very crude hand calculation tells is is a few very interesting things. First, titanium is ridiculously strong, like the safety factors we are going to get with it are going to be insanely high. I need to redo the math on the second page - I'm definitely missing something I don't believe a 1mil titanium tube will hold 1144N of force (then again even if it does axially we haven't calculated any buckling yet). But there's some even more interesting things.


You see how the top bar of the rocket-supports is deflecting the most in the FEA results? These hand calcs actually tell us why that is. The bar that takes the most load in this structure is bar 2 and then the top bar, the bottom bar does almost nothing (again in this very simplified and specific load case). Like 5N? Excuse me? So we need a better design there. That's also why the plate and the frame is bending in the FEA. It's taking the force, not the lower bars those are literally for decoration from what I can see.


Anyways this was a fun experiment. I want to continue this with buckling hand calc and etc to give us better ideas on tube diameters/thicknesses. I have a feeling that's what will give us the most results/answers - well that and the fatigue.


BUT THIS IS A CLASSIC EXAMPLE OF HOW FEA DOES NOT GIVE YOU THE WHOLE STORY --> SO WE'RE GOING TO LITERALLY GO BACK TO THE DRAWING BOARD AND WRITE DOWN THE DESIGN PARAMETERS AND RE-DEVELOP THE STRUCTURE IN 2D SO ALL THE BARS SHARE THE LOAD.


#updates #math #learning

2 views0 comments
© copyright 2019 | aditya mehrotra